add -R to keep only docs with resolved names

1 files changed, 20 insertions, 9 deletions

diff --git a/skate/cmd/skate-resolve-journal-name/main.go b/skate/cmd/skate-resolve-journal-name/main.go
index 2351730..d98f166 100644
--- a/skate/cmd/skate-resolve-journal-name/main.go
+++ b/skate/cmd/skate-resolve-journal-name/main.go
@@ -25,12 +25,13 @@ import (
 )
 
 var (
-	numWorkers = flag.Int("w", runtime.NumCPU(), "number of workers")
-	batchSize  = flag.Int("b", 100000, "batch size")
-	column     = flag.Int("f", 2, "column to find the release schema document in (1-indexed)")
-	sep        = flag.String("d", "\t", "delimiter to use")
-	bestEffort = flag.Bool("B", false, "only log errors, do not halt")
-	abbrevFile = flag.String("A", "", "path to abbreviate file")
+	numWorkers       = flag.Int("w", runtime.NumCPU(), "number of workers")
+	batchSize        = flag.Int("b", 100000, "batch size")
+	column           = flag.Int("f", 2, "column to find the release schema document in (1-indexed)")
+	sep              = flag.String("d", "\t", "delimiter to use")
+	bestEffort       = flag.Bool("B", false, "only log errors, do not halt")
+	abbrevFile       = flag.String("A", "", "path to abbreviate file")
+	keepResolvedOnly = flag.Bool("R", false, "keep only lines, where we could resolve an abbreviation")
 )
 
 func main() {
@@ -60,12 +61,21 @@ func main() {
 		results := ms.Lookup(name, -1)
 		if len(results) != 1 {
 			// To many or too few matches? We return the document unchanged
-			return p, nil
+			if *keepResolvedOnly {
+				return nil, nil
+			} else {
+				return p, nil
+			}
 		}
 		resolved := m[results[0]]
 		if len(resolved) != 1 {
-			// Abbreviation mapping to different full names? Skip.
-			return p, nil
+			// Abbreviation mapping to different full names? Not sure how
+			// common, or why, but skip.
+			if *keepResolvedOnly {
+				return nil, nil
+			} else {
+				return p, nil
+			}
 		}
 		release.Extra.Skate.ResolvedContainerName = resolved[0]
 		b, err := json.Marshal(release)
@@ -73,6 +83,7 @@ func main() {
 			return nil, err
 		}
 		if len(fields) == *column {
+			// In case this is the last field, we want our newline back.
 			b = append(b, []byte("\n")...)
 		}
 		fields[*column-1] = b